1. Field of the Invention
The present invention relates to a communication system for building speech databases for use in speech synthesis, to a relay device therefor, and to a relay method therefor. In particular, the present invention relates to a communication system for building, based on spoken dialogue in telephone and videophone calls, a speech database for use in speech synthesis that focuses on the reproduction of individual characteristics, to a relay device therefor, and to a relay method therefor.
2. Description of Related Art
Speech synthesis technology has been developed with a focus on the naturalness of synthesized speech and individuality so that it is likely that the synthesized speech will be similar to the speech of a human subject.
In such speech synthesis technology, pieces of speech data for a human subject are registered in advance in a database, which was created by recording different pieces of speech of the human subject by causing the human subject to read aloud different stories, and pieces that best match input texts are combined to produce synthesized speech, for example, as described in Japanese Patent Application Laid-Open Publication No. 2003-295880.
However, in the conventional speech synthesis technology, it usually takes many hours of recoding (for example, several to several tens of hours) at a dedicated studio to build a database in which many pieces of speech data for speech synthesis are stored. Therefore, conventional systems can be used for systems that require only limited types of speech patterns, such as a car navigation system or an IVR (Interactive Voice Response) system, but were not suited to reproducing the speech of the human subject in a system such as a mobile communication system.